numbers to guide conservation.
In an article published in the journal Calcutta Statistical Association Bulletin, Dr. Soumen Dey, Prof. Mohan Delampady, Dr. K. Ullas Karanth, and Dr. Arjun M. Gopalaswamy present a new statistical model that increases the accuracy of counting individual tigers in the wild.
Camera traps, devices that automatically trigger upon presence of wildlife along forest trails, are extensively used in the censuses of tigers and other wildlife. Yet, camera traps do not work perfectly. When a pair of camera traps photograph both flanks of a tiger, that individual is considered to be fully identified. However, when only one flank of the tiger is captured due to camera malfunction, it is identified “partially”. This may lead to biases in population counts.
The problem is similar to finding the correct surname of a person in a telephone directory, when the only information available to you is their first name. The new model takes such “missing” data points into account.
Prof. Mohan Delampady, a Senior Professor at ISIBC, says: “Two interesting questions in statistical theory that we faced in this analysis were: How to probabilistically match the left and right flanks of a partially identified tiger, and how to distinguish between a tiger arriving at a camera trap from a tiger arriving, but failing, to get photographed. Such practical, field problems provide motivating ideas in statistical theory.”
The study uses camera trap images obtained by Dr. K. Ullas Karanth, Director at CWS, during his research in Nagarhole National Park in Karnataka, India, from November 2015 to January 2016. He says: “I am pleased that data from my 30-year intensive field study of tigers—conducted in collaboration with Wildlife Conservation Society (WCS), New York—is now helping cutting-edge statistical developments of this kind. I will continue to provide support to such innovation in the years to come.”
“For the dataset we considered, our method improves the precision in tiger density over a previously developed method by 16% and also estimates detector efficiency,” says Dr. Soumen Dey, a researcher at ISIBC. “Our model also potentially finds use in wildlife surveys operating on fixed budgets to optimize the allocation of the number of trap stations and the types of detectors.”
The authors conclude that this study will help ecologists and conservationists address the global challenge of counting and understanding rare and elusive animals.
“Our statistical development not only improves the accuracy of wildlife abundance estimation, but can be used to answer exciting ecological questions, says Dr. Arjun Gopalaswamy, a Visiting Scientist at ISIBC. “For example, we can get answers about why certain locations in forests become more attractive to tigers compared to others. I see a wide application of this statistical model for studies of many iconic species such as tigers, snow leopards, lions and even elephants.”
The study titled “A Spatially Explicit Capture–Recapture Model for Partially Identified Individuals When Trap Detection Rate Is Less than One” authored by Dr. Soumen Dey, Prof. Mohan Delampady, Dr. K. Ullas Karanth, and Dr. Arjun M. Gopalaswamy appeared in the journal Calcutta Statistical Association Bulletin.
